/*
* Copyright (C) 2023 Linux Studio Plugins Project
* (C) 2023 Vladimir Sadovnikov
*
* This file is part of lsp-dsp-lib
* Created on: 2 нояб. 2023 г.
*
* lsp-dsp-lib is free software: you can redistribute it and/or modify
* it under the terms of the GNU Lesser General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* any later version.
*
* lsp-dsp-lib is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with lsp-dsp-lib. If not, see .
*/
#ifndef PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_EXPANDER_H_
#define PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_EXPANDER_H_
#ifndef PRIVATE_DSP_ARCH_X86_AVX512_IMPL
#error "This header should not be included directly"
#endif /* PRIVATE_DSP_ARCH_X86_AVX512_IMPL */
#include
#include
namespace lsp
{
namespace avx512
{
#pragma pack(push, 1)
typedef struct expander_knee_t
{
float start[16]; // +0x000
float end[16]; // +0x040
float threshold[16]; // +0x080
float herm[48]; // +0x0c0
float tilt[32]; // +0x180
} expander_knee_t;
#pragma pack(pop)
static const uint32_t expander_const[] __lsp_aligned64 =
{
LSP_DSP_VEC16(0x7fffffff), // abs
LSP_DSP_VEC16(0x3f800000), // 1.0f
};
#define UNPACK_EXP_KNEE(DST, SRC) \
__ASM_EMIT("vbroadcastss 0x00(%[" SRC "]), %%zmm0") \
__ASM_EMIT("vbroadcastss 0x04(%[" SRC "]), %%zmm1") \
__ASM_EMIT("vbroadcastss 0x08(%[" SRC "]), %%zmm2") \
__ASM_EMIT("vbroadcastss 0x0c(%[" SRC "]), %%zmm3") \
__ASM_EMIT("vbroadcastss 0x10(%[" SRC "]), %%zmm4") \
__ASM_EMIT("vbroadcastss 0x14(%[" SRC "]), %%zmm5") \
__ASM_EMIT("vbroadcastss 0x18(%[" SRC "]), %%zmm6") \
__ASM_EMIT("vbroadcastss 0x1c(%[" SRC "]), %%zmm7") \
__ASM_EMIT("vmovaps %%zmm0, 0x000 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm1, 0x040 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm2, 0x080 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm3, 0x0c0 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm4, 0x100 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm5, 0x140 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm6, 0x180 + %[" DST "]") \
__ASM_EMIT("vmovaps %%zmm7, 0x1c0 + %[" DST "]")
#define PROCESS_UKNEE_SINGLE_X32 \
/* in: zmm0 = lx0, zmm4 = lx1 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm4, %%zmm5") \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm0, %%zmm2") /* zmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm4, %%zmm6") \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm3") /* zmm3 = x0 */ \
__ASM_EMIT("vmovaps 0x040 + %[mem], %%zmm7") \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm1, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm5, %%zmm5") \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm2, %%zmm2") /* zmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm6, %%zmm6") \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm3, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm7, %%k6") \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm5, %%zmm4") \
__ASM_EMIT("vmovaps %%zmm2, %%zmm0 %{%%k4%}") /* zmm0 = [x0 >= end] ? TV : KV */ \
__ASM_EMIT("vmovaps %%zmm6, %%zmm4 %{%%k6%}") \
EXP_CORE_X32 /* zmm0 = EV = expf([x0 >= end] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \
__ASM_EMIT("vmovaps 0x040 + %[mem], %%zmm5") \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm1, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm5, %%k6") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? 1.0 : EV */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm4 %{%%k6%}") \
/* out: zmm0 = g0, zmm4 = g1 */
#define PROCESS_UKNEE_SINGLE_X16 \
/* in: zmm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm0, %%zmm2") /* zmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm3") /* zmm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm1, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm2, %%zmm2") /* zmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm3, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%zmm2, %%zmm0 %{%%k4%}") /* zmm0 = [x0 >= end] ? TV : KV */ \
EXP_CORE_X16 /* zmm0 = EV = expf([x0 >= end] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm1, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? 1.0 : EV */ \
/* out: zmm0 = g0 */
#define PROCESS_UKNEE_SINGLE_X8 \
/* in: ymm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%ymm0, %%ymm1") /* ymm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%ymm0, %%ymm2") /* ymm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%ymm3") /* ymm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%ymm1, %%ymm1") /* ymm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%ymm2, %%ymm2") /* ymm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%ymm3, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%ymm1, %%ymm0") /* ymm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%ymm2, %%ymm0 %{%%k4%}") /* ymm0 = [x0 >= end] ? TV : KV */ \
EXP_CORE_X8 /* ymm0 = EV = expf([x0 >= end] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%ymm1") /* ymm1 = x0 */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%ymm1, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%ymm0 %{%%k4%}") /* ymm0 = [x0 <= start] ? 1.0 : EV */ \
/* out: ymm0 = g0 */
#define PROCESS_UKNEE_SINGLE_X4 \
/* in: xmm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%xmm0, %%xmm1") /* xmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%xmm0, %%xmm2") /* xmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%xmm3") /* xmm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%xmm1, %%xmm1") /* xmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%xmm2, %%xmm2") /* xmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%xmm3, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%xmm1, %%xmm0") /* xmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%xmm2, %%xmm0 %{%%k4%}") /* xmm0 = [x0 >= end] ? TV : KV */ \
EXP_CORE_X4 /* xmm0 = EV = expf([x0 >= end] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%xmm1") /* xmm1 = x0 */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%xmm1, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%xmm0 %{%%k4%}") /* xmm0 = [x0 <= start] ? 1.0 : EV */ \
/* out: xmm0 = g0 */
#define PROCESS_UEXP_FULL_X32 \
/* in: zmm0 = x0, zmm4 = x1 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%zmm4, %%zmm4") \
__ASM_EMIT("vminps 0x080 + %[knee], %%zmm0, %%zmm0") /* zmm0 = min(fabsf(x0), threshold) */ \
__ASM_EMIT("vminps 0x080 + %[knee], %%zmm4, %%zmm4") \
__ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm0, %%k4") /* k4 = [x0 > start] */ \
__ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm4, %%k6") \
__ASM_EMIT("korw %%k6, %%k4, %%k4") \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0") /* zmm0 = 1.0f */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm4") \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vmovaps %%zmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
__ASM_EMIT("vmovaps %%zmm4, 0x040 + %[mem]") \
LOGE_CORE_X32 /* ymm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_UKNEE_SINGLE_X32 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: zmm0 = G0, zmm4 = G1 */
#define PROCESS_UEXP_FULL_X16 \
/* in: zmm0 = x0 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \
__ASM_EMIT("vminps 0x080 + %[knee], %%zmm0, %%zmm0") /* zmm0 = min(fabsf(x0), threshold) */ \
__ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm0, %%k4") /* k4 = [x0 > start] */ \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0") /* zmm0 = 1.0f */ \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vmovaps %%zmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X16 /* zmm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_UKNEE_SINGLE_X16 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: zmm0 = G0 */
#define PROCESS_UEXP_FULL_X8 \
/* in: ymm0 = x0 */ \
__ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%ymm0, %%k4") /* k4 = [x0 > start] */ \
__ASM_EMIT("vminps 0x080 + %[knee], %%ymm0, %%ymm0") /* ymm0 = min(fabsf(x0), threshold) */ \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%ymm0") /* ymm0 = 1.0f */ \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vandps 0x00 + %[X2C], %%ymm0, %%ymm0") /* ymm0 = fabsf(x0) */ \
__ASM_EMIT("vmovaps %%ymm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X8 /* ymm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_UKNEE_SINGLE_X8 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: ymm0 = G0 */
#define PROCESS_UEXP_FULL_X4 \
/* in: xmm0 = x0 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%xmm0, %%xmm0") /* xmm0 = fabsf(x0) */ \
__ASM_EMIT("vminps 0x080 + %[knee], %%xmm0, %%xmm0") /* xmm0 = min(fabsf(x0), threshold) */ \
__ASM_EMIT("vmovaps %%xmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X4 /* xmm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_UKNEE_SINGLE_X4 /* apply knee */ \
/* out: xmm0 = G0 */
void uexpander_x1_gain(float *dst, const float *src, const dsp::expander_knee_t *c, size_t count)
{
IF_ARCH_X86(
expander_knee_t knee __lsp_aligned64;
float mem[64] __lsp_aligned64;
float stub[16] __lsp_aligned64;
size_t mask;
);
ARCH_X86_ASM
(
// Prepare stuff
UNPACK_EXP_KNEE("knee", "exp")
// 32x blocks
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jb 2f")
__ASM_EMIT("1:")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
__ASM_EMIT("vmovups 0x40(%[src]), %%zmm4")
PROCESS_UEXP_FULL_X32
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])")
__ASM_EMIT("add $0x80, %[src]")
__ASM_EMIT("add $0x80, %[dst]")
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jae 1b")
__ASM_EMIT("2:")
// 16x blocks
__ASM_EMIT("add $16, %[count]")
__ASM_EMIT("jl 4f")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
PROCESS_UEXP_FULL_X16
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("add $0x40, %[src]")
__ASM_EMIT("add $0x40, %[dst]")
__ASM_EMIT("sub $16, %[count]")
__ASM_EMIT("4:")
// 8x blocks
__ASM_EMIT("add $8, %[count]")
__ASM_EMIT("jl 6f")
__ASM_EMIT("vmovups 0x00(%[src]), %%ymm0")
PROCESS_UEXP_FULL_X8
__ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])")
__ASM_EMIT("add $0x20, %[src]")
__ASM_EMIT("add $0x20, %[dst]")
__ASM_EMIT("sub $8, %[count]")
__ASM_EMIT("6:")
// 4x blocks
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jl 8f")
__ASM_EMIT("vmovups 0x00(%[src]), %%xmm0")
PROCESS_UEXP_FULL_X4
__ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])")
__ASM_EMIT("sub $4, %[count]")
__ASM_EMIT("add $0x10, %[src]")
__ASM_EMIT("add $0x10, %[dst]")
__ASM_EMIT("8:")
// Tail: 1x-3x block
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jle 16f")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 10f")
__ASM_EMIT("vmovss 0x00(%[src]), %%xmm0")
__ASM_EMIT("add $4, %[src]")
__ASM_EMIT("10:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 12f")
__ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("12:")
PROCESS_UEXP_FULL_X4
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 14f")
__ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])")
__ASM_EMIT("add $4, %[dst]")
__ASM_EMIT("14:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 16f")
__ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])")
__ASM_EMIT("16:")
: [dst] "+r" (dst), [src] "+r" (src),
[count] "+r" (count),
[mask] "=&r" (mask)
: [exp] "r" (c),
[knee] "o" (knee),
[mem] "o" (mem),
[stub] "o" (stub),
[X2C] "o" (expander_const),
[L2C] "o" (LOG2_CONST),
[LOGC] "o" (LOGE_C),
[E2C] "o" (EXP2_CONST),
[LOG2E] "m" (EXP_LOG2E)
: "cc", "memory",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%k4", "%k5", "%k6", "%k7"
);
}
void uexpander_x1_curve(float *dst, const float *src, const dsp::expander_knee_t *c, size_t count)
{
IF_ARCH_X86(
expander_knee_t knee __lsp_aligned64;
float mem[64] __lsp_aligned64;
float stub[16] __lsp_aligned64;
size_t mask;
);
ARCH_X86_ASM
(
// Prepare stuff
UNPACK_EXP_KNEE("knee", "exp")
// 32x blocks
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jb 2f")
__ASM_EMIT("1:")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
__ASM_EMIT("vmovups 0x40(%[src]), %%zmm4")
PROCESS_UEXP_FULL_X32
__ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0")
__ASM_EMIT("vmulps 0x40(%[src]), %%zmm4, %%zmm4")
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])")
__ASM_EMIT("add $0x80, %[src]")
__ASM_EMIT("add $0x80, %[dst]")
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jae 1b")
__ASM_EMIT("2:")
// 16x blocks
__ASM_EMIT("add $16, %[count]")
__ASM_EMIT("jl 4f")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
PROCESS_UEXP_FULL_X16
__ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0")
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("add $0x40, %[src]")
__ASM_EMIT("add $0x40, %[dst]")
__ASM_EMIT("sub $16, %[count]")
__ASM_EMIT("4:")
// 8x blocks
__ASM_EMIT("add $8, %[count]")
__ASM_EMIT("jl 6f")
__ASM_EMIT("vmovups 0x00(%[src]), %%ymm0")
PROCESS_UEXP_FULL_X8
__ASM_EMIT("vmulps 0x00(%[src]), %%ymm0, %%ymm0")
__ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])")
__ASM_EMIT("add $0x20, %[src]")
__ASM_EMIT("add $0x20, %[dst]")
__ASM_EMIT("sub $8, %[count]")
__ASM_EMIT("6:")
// 4x blocks
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jl 8f")
__ASM_EMIT("vmovups 0x00(%[src]), %%xmm0")
PROCESS_UEXP_FULL_X4
__ASM_EMIT("vmulps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])")
__ASM_EMIT("sub $4, %[count]")
__ASM_EMIT("add $0x10, %[src]")
__ASM_EMIT("add $0x10, %[dst]")
__ASM_EMIT("8:")
// Tail: 1x-3x block
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jle 16f")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 10f")
__ASM_EMIT("vmovss 0x00(%[src]), %%xmm0")
__ASM_EMIT("add $4, %[src]")
__ASM_EMIT("10:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 12f")
__ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("12:")
__ASM_EMIT("vmovaps %%xmm0, %%xmm4")
PROCESS_UEXP_FULL_X4
__ASM_EMIT("vmulps %%xmm4, %%xmm0, %%xmm0")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 14f")
__ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])")
__ASM_EMIT("add $4, %[dst]")
__ASM_EMIT("14:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 16f")
__ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])")
__ASM_EMIT("16:")
: [dst] "+r" (dst), [src] "+r" (src),
[count] "+r" (count),
[mask] "=&r" (mask)
: [exp] "r" (c),
[knee] "o" (knee),
[mem] "o" (mem),
[stub] "o" (stub),
[X2C] "o" (expander_const),
[L2C] "o" (LOG2_CONST),
[LOGC] "o" (LOGE_C),
[E2C] "o" (EXP2_CONST),
[LOG2E] "m" (EXP_LOG2E)
: "cc", "memory",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%k4", "%k5", "%k6", "%k7"
);
}
#undef PROCESS_UEXP_FULL_X4
#undef PROCESS_UEXP_FULL_X8
#undef PROCESS_UEXP_FULL_X16
#undef PROCESS_UEXP_FULL_X32
#undef PROCESS_UKNEE_SINGLE_X4
#undef PROCESS_UKNEE_SINGLE_X8
#undef PROCESS_UKNEE_SINGLE_X16
#undef PROCESS_UKNEE_SINGLE_X32
#define PROCESS_DKNEE_SINGLE_X32 \
/* in: ymm0 = lx0, ymm4 = lx1, ymm8 = lx2, ymm12 = lx3 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm4, %%zmm5") \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm0, %%zmm2") /* zmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm4, %%zmm6") \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm3") /* zmm3 = x0 */ \
__ASM_EMIT("vmovaps 0x040 + %[mem], %%zmm7") \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm1, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm5, %%zmm5") \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm2, %%zmm2") /* zmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm6, %%zmm6") \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm3, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm7, %%k6") \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm5, %%zmm4") \
__ASM_EMIT("vmovaps %%zmm2, %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? TV : KV */ \
__ASM_EMIT("vmovaps %%zmm6, %%zmm4 %{%%k6%}") \
EXP_CORE_X32 /* zmm0 = EV = expf([x0 <= start] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \
__ASM_EMIT("vmovaps 0x040 + %[mem], %%zmm5") \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm1, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm5, %%k6") \
__ASM_EMIT("vcmpps $1, 0x080 + %[knee], %%zmm1, %%k5") /* k5 = [x0 < threshold] */ \
__ASM_EMIT("vcmpps $1, 0x080 + %[knee], %%zmm5, %%k7") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0 %{%%k4%}") /* zmm0 = [x0 >= end] ? 1.0 : EV */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm4 %{%%k6%}") \
__ASM_EMIT("vxorps %%zmm0, %%zmm0, %%zmm0 %{%%k5%}") /* zmm0 = [x0 < threshold] ? 0 : [x0 >= end] ? 1.0 : EV */ \
__ASM_EMIT("vxorps %%zmm4, %%zmm4, %%zmm4 %{%%k7%}") \
/* out: ymm0 = g0, ymm4 = g1, ymm8 = g2, ymm12 = g3 */
#define PROCESS_DKNEE_SINGLE_X16 \
/* in: zmm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%zmm0, %%zmm2") /* zmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm3") /* zmm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%zmm1, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%zmm2, %%zmm2") /* zmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%zmm3, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%zmm2, %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? TV : KV */ \
EXP_CORE_X16 /* zmm0 = EV = expf([x0 <= start] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%zmm1, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vcmpps $1, 0x080 + %[knee], %%zmm1, %%k5") /* k5 = [x0 < threshold] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0 %{%%k4%}") /* zmm0 = [x0 >= end] ? 1.0 : EV */ \
__ASM_EMIT("vxorps %%zmm0, %%zmm0, %%zmm0 %{%%k5%}") /* zmm0 = [x0 < threshold] ? 0 : [x0 >= end] ? 1.0 : EV */ \
/* out: zmm0 = g0 */
#define PROCESS_DKNEE_SINGLE_X8 \
/* in: ymm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%ymm0, %%ymm1") /* ymm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%ymm0, %%ymm2") /* ymm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%ymm3") /* ymm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%ymm1, %%ymm1") /* ymm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%ymm2, %%ymm2") /* ymm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%ymm3, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%ymm1, %%ymm0") /* ymm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%ymm2, %%ymm0 %{%%k4%}") /* ymm0 = [x0 <= start] ? TV : KV */ \
EXP_CORE_X8 /* ymm0 = EV = expf([x0 <= start] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%ymm1") /* ymm1 = x0 */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%ymm1, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vcmpps $1, 0x080 + %[knee], %%ymm1, %%k5") /* k5 = [x0 < threshold] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%ymm0 %{%%k4%}") /* ymm0 = [x0 >= end] ? 1.0 : EV */ \
__ASM_EMIT("vxorps %%ymm0, %%ymm0, %%ymm0 %{%%k5%}") /* ymm0 = [x0 < threshold] ? 0 : [x0 >= end] ? 1.0 : EV */ \
/* out: ymm0 = g0 */
#define PROCESS_DKNEE_SINGLE_X4 \
/* in: xmm0 = lx0 */ \
__ASM_EMIT("vmulps 0x0c0 + %[knee], %%xmm0, %%xmm1") /* xmm1 = herm[0]*lx0 */ \
__ASM_EMIT("vmulps 0x180 + %[knee], %%xmm0, %%xmm2") /* xmm2 = tilt[0]*lx0 */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%xmm3") /* xmm3 = x0 */ \
__ASM_EMIT("vaddps 0x100 + %[knee], %%xmm1, %%xmm1") /* xmm1 = herm[0]*lx0+herm[1] */ \
__ASM_EMIT("vaddps 0x1c0 + %[knee], %%xmm2, %%xmm2") /* xmm2 = TV = tilt[0]*lx0+tilt[1] */ \
__ASM_EMIT("vcmpps $2, 0x000 + %[knee], %%xmm3, %%k4") /* k4 = [x0 <= start] */ \
__ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%xmm1, %%xmm0") /* xmm0 = KV = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \
__ASM_EMIT("vmovaps %%xmm2, %%xmm0 %{%%k4%}") /* xmm0 = [x0 <= start] ? TV : KV */ \
EXP_CORE_X4 /* xmm0 = EV = expf([x0 <= start] ? TV : KV) */ \
__ASM_EMIT("vmovaps 0x000 + %[mem], %%xmm1") /* xmm1 = x0 */ \
__ASM_EMIT("vcmpps $5, 0x040 + %[knee], %%xmm1, %%k4") /* k4 = [x0 >= end] */ \
__ASM_EMIT("vcmpps $1, 0x080 + %[knee], %%xmm1, %%k5") /* k5 = [x0 < threshold] */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%xmm0 %{%%k4%}") /* xmm0 = [x0 >= end] ? 1.0 : EV */ \
__ASM_EMIT("vxorps %%xmm0, %%xmm0, %%xmm0 %{%%k5%}") /* xmm0 = [x0 < threshold] ? 0 : [x0 >= end] ? 1.0 : EV */ \
/* out: xmm0 = g0 */
#define PROCESS_DEXP_FULL_X32 \
/* in: ymm0 = x0, ymm4 = x1, ymm8 = x2, ymm12 = x4 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \
__ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm0, %%k4") /* k4 = [x0 < end] */ \
__ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm4, %%k6") \
__ASM_EMIT("korw %%k6, %%k4, %%k4") \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0") /* zmm0 = 1.0f */ \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm4") \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vmovaps %%zmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
__ASM_EMIT("vmovaps %%zmm4, 0x040 + %[mem]") \
LOGE_CORE_X32 /* zmm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_DKNEE_SINGLE_X32 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: zmm0 = G0, zmm4 = G1, zmm8 = G2, zmm12 = G3 */
#define PROCESS_DEXP_FULL_X16 \
/* in: zmm0 = x0 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \
__ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm0, %%k4") /* k4 = [x0 < end] */ \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%zmm0") /* zmm0 = 1.0f */ \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vmovaps %%zmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X16 /* ymm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_DKNEE_SINGLE_X16 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: zmm0 = G0 */
#define PROCESS_DEXP_FULL_X8 \
/* in: ymm0 = x0 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%ymm0, %%ymm0") /* ymm0 = fabsf(x0) */ \
__ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%ymm0, %%k4") /* k4 = [x0 < end] */ \
__ASM_EMIT("kmovw %%k4, %k[mask]") \
__ASM_EMIT("test %k[mask], %k[mask]") \
__ASM_EMIT("jnz 100f") \
__ASM_EMIT("vmovaps 0x040 + %[X2C], %%ymm0") /* ymm0 = 1.0f */ \
__ASM_EMIT("jmp 200f") \
__ASM_EMIT("100:") \
__ASM_EMIT("vmovaps %%ymm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X8 /* ymm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_DKNEE_SINGLE_X8 /* apply knee */ \
__ASM_EMIT("200:") \
/* out: ymm0 = G0 */
#define PROCESS_DEXP_FULL_X4 \
/* in: xmm0 = x0 */ \
__ASM_EMIT("vandps 0x000 + %[X2C], %%xmm0, %%xmm0") /* xmm0 = fabsf(x0) */ \
__ASM_EMIT("vmovaps %%xmm0, 0x000 + %[mem]") /* store fabsf(x0) */ \
LOGE_CORE_X4 /* xmm0 = lx0 = logf(fabsf(x0)) */ \
PROCESS_DKNEE_SINGLE_X4 /* apply knee */ \
/* out: xmm0 = G0 */
void dexpander_x1_gain(float *dst, const float *src, const dsp::expander_knee_t *c, size_t count)
{
IF_ARCH_X86(
expander_knee_t knee __lsp_aligned64;
float mem[64] __lsp_aligned64;
float stub[16] __lsp_aligned64;
size_t mask;
);
ARCH_X86_ASM
(
// Prepare stuff
UNPACK_EXP_KNEE("knee", "exp")
// 32x blocks
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jb 2f")
__ASM_EMIT("1:")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
__ASM_EMIT("vmovups 0x40(%[src]), %%zmm4")
PROCESS_DEXP_FULL_X32
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])")
__ASM_EMIT("add $0x80, %[src]")
__ASM_EMIT("add $0x80, %[dst]")
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jae 1b")
__ASM_EMIT("2:")
// 16x blocks
__ASM_EMIT("add $16, %[count]")
__ASM_EMIT("jl 4f")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
PROCESS_DEXP_FULL_X16
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("add $0x40, %[src]")
__ASM_EMIT("add $0x40, %[dst]")
__ASM_EMIT("sub $16, %[count]")
__ASM_EMIT("4:")
// 8x blocks
__ASM_EMIT("add $8, %[count]")
__ASM_EMIT("jl 6f")
__ASM_EMIT("vmovups 0x00(%[src]), %%ymm0")
PROCESS_DEXP_FULL_X8
__ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])")
__ASM_EMIT("add $0x20, %[src]")
__ASM_EMIT("add $0x20, %[dst]")
__ASM_EMIT("sub $8, %[count]")
__ASM_EMIT("6:")
// 4x blocks
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jl 8f")
__ASM_EMIT("vmovups 0x00(%[src]), %%xmm0")
PROCESS_DEXP_FULL_X4
__ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])")
__ASM_EMIT("sub $4, %[count]")
__ASM_EMIT("add $0x10, %[src]")
__ASM_EMIT("add $0x10, %[dst]")
__ASM_EMIT("8:")
// Tail: 1x-3x block
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jle 16f")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 10f")
__ASM_EMIT("vmovss 0x00(%[src]), %%xmm0")
__ASM_EMIT("add $4, %[src]")
__ASM_EMIT("10:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 12f")
__ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("12:")
PROCESS_DEXP_FULL_X4
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 14f")
__ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])")
__ASM_EMIT("add $4, %[dst]")
__ASM_EMIT("14:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 16f")
__ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])")
__ASM_EMIT("16:")
: [dst] "+r" (dst), [src] "+r" (src),
[count] "+r" (count),
[mask] "=&r" (mask)
: [exp] "r" (c),
[knee] "o" (knee),
[mem] "o" (mem),
[stub] "o" (stub),
[X2C] "o" (expander_const),
[L2C] "o" (LOG2_CONST),
[LOGC] "o" (LOGE_C),
[E2C] "o" (EXP2_CONST),
[LOG2E] "m" (EXP_LOG2E)
: "cc", "memory",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%k4", "%k5", "%k6", "%k7"
);
}
void dexpander_x1_curve(float *dst, const float *src, const dsp::expander_knee_t *c, size_t count)
{
IF_ARCH_X86(
expander_knee_t knee __lsp_aligned64;
float mem[64] __lsp_aligned64;
float stub[16] __lsp_aligned64;
size_t mask;
);
ARCH_X86_ASM
(
// Prepare stuff
UNPACK_EXP_KNEE("knee", "exp")
// 32x blocks
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jb 2f")
__ASM_EMIT("1:")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
__ASM_EMIT("vmovups 0x40(%[src]), %%zmm4")
PROCESS_DEXP_FULL_X32
__ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0")
__ASM_EMIT("vmulps 0x40(%[src]), %%zmm4, %%zmm4")
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])")
__ASM_EMIT("add $0x80, %[src]")
__ASM_EMIT("add $0x80, %[dst]")
__ASM_EMIT("sub $32, %[count]")
__ASM_EMIT("jae 1b")
__ASM_EMIT("2:")
// 16x blocks
__ASM_EMIT("add $16, %[count]")
__ASM_EMIT("jl 4f")
__ASM_EMIT("vmovups 0x00(%[src]), %%zmm0")
PROCESS_DEXP_FULL_X16
__ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0")
__ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])")
__ASM_EMIT("add $0x40, %[src]")
__ASM_EMIT("add $0x40, %[dst]")
__ASM_EMIT("sub $16, %[count]")
__ASM_EMIT("4:")
// 8x blocks
__ASM_EMIT("add $8, %[count]")
__ASM_EMIT("jl 6f")
__ASM_EMIT("vmovups 0x00(%[src]), %%ymm0")
PROCESS_DEXP_FULL_X8
__ASM_EMIT("vmulps 0x00(%[src]), %%ymm0, %%ymm0")
__ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])")
__ASM_EMIT("add $0x20, %[src]")
__ASM_EMIT("add $0x20, %[dst]")
__ASM_EMIT("sub $8, %[count]")
__ASM_EMIT("6:")
// 4x blocks
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jl 8f")
__ASM_EMIT("vmovups 0x00(%[src]), %%xmm0")
PROCESS_DEXP_FULL_X4
__ASM_EMIT("vmulps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])")
__ASM_EMIT("sub $4, %[count]")
__ASM_EMIT("add $0x10, %[src]")
__ASM_EMIT("add $0x10, %[dst]")
__ASM_EMIT("8:")
// Tail: 1x-3x block
__ASM_EMIT("add $4, %[count]")
__ASM_EMIT("jle 16f")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 10f")
__ASM_EMIT("vmovss 0x00(%[src]), %%xmm0")
__ASM_EMIT("add $4, %[src]")
__ASM_EMIT("10:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 12f")
__ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0")
__ASM_EMIT("12:")
__ASM_EMIT("vmovaps %%xmm0, %%xmm4")
PROCESS_DEXP_FULL_X4
__ASM_EMIT("vmulps %%xmm4, %%xmm0, %%xmm0")
__ASM_EMIT("test $1, %[count]")
__ASM_EMIT("jz 14f")
__ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])")
__ASM_EMIT("add $4, %[dst]")
__ASM_EMIT("14:")
__ASM_EMIT("test $2, %[count]")
__ASM_EMIT("jz 16f")
__ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])")
__ASM_EMIT("16:")
: [dst] "+r" (dst), [src] "+r" (src),
[count] "+r" (count),
[mask] "=&r" (mask)
: [exp] "r" (c),
[knee] "o" (knee),
[mem] "o" (mem),
[stub] "o" (stub),
[X2C] "o" (expander_const),
[L2C] "o" (LOG2_CONST),
[LOGC] "o" (LOGE_C),
[E2C] "o" (EXP2_CONST),
[LOG2E] "m" (EXP_LOG2E)
: "cc", "memory",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%k4", "%k5", "%k6", "%k7"
);
}
#undef PROCESS_DEXP_FULL_X4
#undef PROCESS_DEXP_FULL_X8
#undef PROCESS_DEXP_FULL_X16
#undef PROCESS_DEXP_FULL_X32
#undef PROCESS_DKNEE_SINGLE_X4
#undef PROCESS_DKNEE_SINGLE_X8
#undef PROCESS_DKNEE_SINGLE_X16
#undef PROCESS_DKNEE_SINGLE_X32
#undef UNPACK_EXP_KNEE
} /* namespace avx512 */
} /* namespace lsp */
#endif /* PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_EXPANDER_H_ */